Our field isn't quite "artificial intelligence" -- it's "cognitive automation": the encoding and operationalization of human-generated abstractions / behaviors / skills. The "intelligence" label is a category error
Cognitive automation can happen via explicitly hard-coding human-generated rules (GOFAI), or via collecting a dense sampling of labeled inputs and training some flavor of locality-sensitive hashtable on it (such as a deep learning model)
The second form is especially powerful, since encoding implicit abstractions only via labeled training examples is far more practical and versatile than explicitly programming abstractions by hand, for all kinds of historically difficult problems.
Cognitive automation is incredibly useful. But autonomous abstraction generation is a different creature altogether.
As new lifeforms are to animated cartoon characters -- whether the cartoon character is modeled by hand or captured via example
As new lifeforms are to animated cartoon characters -- whether the cartoon character is modeled by hand or captured via example
"If the cartoon is drawn with sufficient realism and covers sufficiently many scenes, what's the difference?", you may ask.
Adaptability to the unknown. A lifeform will autonomously adapt to a changing future. An automaton will perform the scenes you planned for.
Adaptability to the unknown. A lifeform will autonomously adapt to a changing future. An automaton will perform the scenes you planned for.
Intelligence is adaption to unknown unknowns across an unknown range of tasks and domains.
Automation is, at best, robustly handling known unknowns over known tasks (which is already incredibly difficult and resource-intensive in the real world -- whether engineering or data)
Automation is, at best, robustly handling known unknowns over known tasks (which is already incredibly difficult and resource-intensive in the real world -- whether engineering or data)
The resource-intensiveness, naturally, comes from the lack of adaptability: you need to plan for every possible unknown, whether explicitly or via a dense sampling of possible situations (assuming a fixed distribution)
